1 |
On 8 December 2014 at 08:54, Rich Freeman <rich0@g.o> wrote: |
2 |
> On Sun, Dec 7, 2014 at 5:42 PM, James <wireless@×××××××××××.com> wrote: |
3 |
>> Rich Freeman <rich0 <at> gentoo.org> writes: |
4 |
>>> > liveUSB, where folks can download "Gentoo Fever" onto a usb stick and stick |
5 |
>>> > into their current hardware and boot up a killer code development system. |
6 |
>>> |
7 |
>>> Building a liveUSB version of Gentoo is almost completely orthagonal |
8 |
>>> to building an automatic ebuild testing system. |
9 |
>> |
10 |
>> Agreed from where you sit. Where I sit, when I explain to folks about |
11 |
>> this project, a liveUSB stick running the latest in what gentoo-fever |
12 |
>> is, speaks volumes to encourage non-gentoo folks to take it for a test |
13 |
>> drive. |
14 |
> |
15 |
> I didn't say it was a bad idea. I just said that it had nothing to do |
16 |
> with Continuous Integration (CI). |
17 |
> |
18 |
> If a billion people used Gentoo for 8 hours a day straight, that would |
19 |
> also have nothing to do with CI. |
20 |
> |
21 |
>> |
22 |
>>> |
23 |
>>> A Gentoo CI system doesn't even have to be hosted on Gentoo, or on |
24 |
>>> Linux for that matter. Of course, if one were ever to become official |
25 |
>>> it most likely would be hosted on Gentoo, but most likely not on a box |
26 |
>>> booted from a USB. |
27 |
>> |
28 |
>> Um, I never saw "CI" defined, so please define specifically, then |
29 |
>> use the abbrev? |
30 |
>> |
31 |
> |
32 |
> Continuous Integration is probably being used a bit loosely here. The |
33 |
> concept is testing every commit to ensure a level of quality. Commits |
34 |
> wouldn't directly hit users - they would be tested first (perhaps in |
35 |
> batches), and then would only hit users if they pass. Or something |
36 |
> along those lines. Maybe it would just refer to frequently testing |
37 |
> the tree with automated bug reporting. |
38 |
> |
39 |
> None of this has to do with having a USB live distro. |
40 |
> |
41 |
|
42 |
Well the same kind of minimal core is useful/needed as the 'base' from which |
43 |
ebuild test runs can be done on top of. So its not a completely |
44 |
orthogonal problem. |
45 |
|
46 |
>> |
47 |
>>> I'm not saying that a liveUSB version of Gentoo wouldn't be nice to |
48 |
>>> have. It just has nothing to do with solving this particular problem. |
49 |
>> |
50 |
>> OK, see above; you are right technically. Do you want a few dozen |
51 |
>> participates or a few thousand? |
52 |
> |
53 |
> I want ZERO participants. That's the whole point. It is supposed to |
54 |
> be automated. |
55 |
> |
56 |
> Sure, somebody has to write the code, but I doubt handing out liveUSB |
57 |
> images is going to inspire that. |
58 |
> |
59 |
> Also, if your main goal was to have a quick easy-to-use Linux desktop, |
60 |
> I'm not quite sure why you'd pick Gentoo in particular to base it on. |
61 |
> The whole point of Gentoo is that you can change it, while a liveUSB |
62 |
> tends to imply something static and standardized. |
63 |
> |
64 |
> Heck, they could use ChromeOS as it is a Gentoo derivative. However, |
65 |
> if you want something in-between Sabayon is probably about right. :) |
66 |
> |
67 |
|
68 |
Gentoo is actually capable of everything that draws people to Arch Linux. |
69 |
The only reason anyone ever gives me for using Arch Linux is the |
70 |
AUR which is pretty much 'portage plus layman for dummies'. So Gentoo |
71 |
does have the ability to give people what they want, we just have no way |
72 |
to support that kind of use case at present. |
73 |
|
74 |
I agree a live USB image wont really 'inspire' anything. But it does represent |
75 |
a useful goal. The same infrastructure that tests ebuilds would be able |
76 |
to generate boot-able images. You may not have been far off the mark |
77 |
with ChromeOS, I think it may be possible to use those tools to build an |
78 |
image without much fuss at all. I would need to remove a fair number of |
79 |
ChromeOS and CoreOS parts from the setup I currently have but it should |
80 |
in theory work fairly easily. But other than improving the process of building |
81 |
a base image, it doesn't really get us anywhere closer to the goal of |
82 |
automation and CI. |
83 |
|
84 |
Rich, you mentioned zero participants and your right on the money, which |
85 |
is why I'm still wondering where people want to collaborate on this effort. |
86 |
Working code and projects is nice but having things spread over several |
87 |
projects doesn't help keep things easy to manage or deal with discussions |
88 |
about how we integrate things together. For instance at some point this |
89 |
should integrate into the ebuild arch unstable masking for all of Gentoo |
90 |
and then things get 'political'. |